The long-term objective of the proposed research program is to reduce smoking-related morbidity and mortality by developing and fielding an effective school-based anti-smoking program. The specific aims are to revise, implement, and evaluate a smoking prevention program which has been successfully pilot-tested. The research plan calls for implementation of the revised program in several urban middle schools and in one small town junior high school. Classrooms will be randomized into either the experimental or the control condition. A Solomon four group design will be used, with pretesting of a randomly selected two thirds of the experimental classrooms, and a randomly selected two thirds of the control classrooms. This will allow an assessment of the generalizability of the program to un-pretested populations, a question of considerable practical importance. A physiological measurement of thiocyanate in either saliva or expired air will be taken on a randomly-selected one fourth of the experimental subjects and one fourth of the control subjects. Repeated posttests at six month intervals continue throughout the life of the study. The smoking prevention program to be tested shares many of the features of those newer programs which have demonstrated some success in (at least) delaying the adoption of cigarette smoking. These features include information on the immediate (as well as long-term) adverse health effects of smoking, suggested social skills with which to resist peer pressure to smoke, active role-playing by subjects as rehearsals for using these social skills, and information on the actual proportion of youngsters and adults who smoke. In addition, the program to be tested provides subjects with a knowledge base to enable them to interpret certain common smoking experiences as signs of bodily damage. The program attributes aversive reactions (coughing, etc.) to the poisons in cigarettes, and also attributes symptom abatement (adaptation) to poisons which have destroyed the body's warning system. The goal is to train smokers to infer injury from either symptoms OR their absence, and to make this inference process automatic. The research design includes measures of process (how well or poorly messages are understood and accepted), and measures of outcome (smoking intentions and behaviors).